Accuracy of visual encodings
Ranks the ways of showing a value (position, length, angle, area, color) by how accurately people read and compare them.
What problem it addresses
It explains why the same data are easy to compare in bars and hard in a pie chart. It gives design a criterion for deciding how to draw each value.
An example
An illustrative example: five products with similar sales, shown in two ways.
The data are the same; what changes is the task you ask the eye to do.
Why it happens
Every chart asks for a visual task: judging a position, a length, an angle or an area;
not all of them are equally accurate, and position along a common axis is the one we read best;
if the key value goes into a hard task, it gets compared poorly even if the chart is correct.
Applications in design
Use it like this
Put what matters most in position along a common axis; leave color and area for secondary information.
Avoid
Asking for fine comparisons with angles, areas or shades of color.
Don’t fall for it
If you need to read the labels to know which value is larger, the encoding isn’t working.
1.96 times
What the research says
Comparing positions in bars was more accurate than comparing angles in a pie chart, in an experiment with 51 people (Cleveland and McGill, 1984).
Solid evidence. An online replication confirmed that position wins, although angle did not come out worse than length, as the ranking predicted (Heer and Bostock, 2010).
Connections
Authors and works
- 1984
William S. Cleveland and Robert McGill
Proposed the ranking of visual tasks and tested it with experiments comparing values.
- 2010
Jeffrey Heer and Michael Bostock
Repeated the experiments with online participants, confirmed the overall ranking and added rectangle areas.
Sources
Original sources
- Cleveland, W. S., & McGill, R. (1984). Graphical perception: Theory, experimentation, and application to the development of graphical methods. Journal of the American Statistical Association, 79(387), 531–554. View source: Cleveland, W. S., & McGill, R. (1984)
- Heer, J., & Bostock, M. (2010). Crowdsourcing graphical perception: Using Mechanical Turk to assess visualization design. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 203–212). ACM. View source: Heer, J., & Bostock, M. (2010)
Last reviewed: